In the preceding section, we have shown that the PCM has the potential to apply to different hypnotic phenomena. However, the literature on hypnosis shows that there are strong inter-individual differences in the level of hypnotic suggestibility. As described in Section 1, suggestions are usually divided into three categories: motor suggestions (e.g., hand lowering, arm levitation, magnetic hands), challenge suggestions (e.g., arm rigidity; the participant is suggested that her arm is unbendable but asked to test this fact by trying to bend it) and cognitive suggestions (e.g., positive and negative hallucinations, amnesia). Now, about 80% of people can pass motor suggestions, about 50% challenge suggestions and 10% cognitive suggestions (see e.g., Hilgard, 1965; Kallio & Ihamuotila, 1999; Perry, Nadon, & Button, 1992).
The existence of these inter-individual differences in hypnotic performances is one of the most intriguing issues in the field of hypnosis and is still in need of a convincing explanation. Many researchers argue that differences in hypnotisability may reflect other trait differences among individuals (But see e.g., Barber, 1969; Sapnos, 1986). Nonetheless, no firm cognitive profiles differentiating highs from lows surface from the literature when tested in non-hypnotic contexts (Council, Kirsch, & Hafner 1986; Heap, Brown, & Oakley 2004;
Laurence, Beaulieu-Prévost, & Du Chéné 2008). In addition, no marker, whether genetic, physiological, behavioural or phenomenological, allowing us to categorize an individual as high, medium or low, has been found so far. In other words, the question of inter-individual differences in hypnotic suggestibility is still unsettled (Martin, Sackur & Dienes, 2017).
An important proposal in the field has been that inter-individual differences in hypnotic suggestibility would result from highs exhibiting more efficient executive functions, such as better sustained and/or selective attention, than lows or mediums (Crawford, 1991, 1994 ; Crawford, Brown & Moon, 1993). The relevant literature shows strong inconsistency, however.
A number of studies found no significant behavioural differences related to subjects’ level of hypnotic suggestibility in baseline performances for various executive and attentional tasks (Cojan, Piguet, & Vuilleumier 2015; Dienes et al. 2009; Egner, Jamieson, & Gruzelier 2005;
Iani, Ricci, Gherri, & Rubichi 2006; Iani, Ricci, Baroni, & Rubichi 2009; Raz, Fan, & Posner 2005; Varga, Németh, & Szekely 2011). Other studies found demonstrate significant differences, but going in either direction (Crawford et al. 1993; Dixon, Brunet, & Laurence 1990; Dixon & Laurence 1992; Farvolden & Woody 2004; Miller, Hen- nessy, & Leibowitz 1973; Martin et al., 2017; Miller 1975; Rubichi, Ricci, Padovani, & Scaglietti 2005; Wallace 1986; Wallace & Garrett 1973; Wallace, Garrett, & Anstadt 1974; Wallace, Knight, & Garrett 1976)
Within the current predictive account of hypnosis, we propose that the difficulty of a suggestion would depend on the level of sensory evidence provided by the suggested event and inter-individual differences would result from inter-individual differences in the “ability” to assign more or less weight to ones’ prior expectations.
The main focus of the present article was one sense of agency alterations with respect to motor suggestions in which inter-individual differences are weak. With motor suggestions, there is a huge amount of sensory evidence as the participant is moving one or another limb so that somesthetic cues (e.g., proprioceptive signals) are changing over time giving a lot of sensory information. Therefore, the available sensory evidence – whose phenomenological saliency is increased by attention – easily fits the prior of non-agency provided by the motor suggestion.
At the opposite end we find cognitive suggestions, such as positive hallucinations, in which there is no distal sensory evidence available. We hypothesise that inter-individual differences might result from inter-individual differences in the “ability” to assign more or less
weight to the perceptual prior. In the preceding section, we proposed that hypnotic hallucinations result from a match between predictions generated by a perceptual prior provided by the suggestion and a mental image with high sensory qualities. However, the level of sensory evidence provided by the mental image might not be sufficient in itself for the image to be mistaken for a perceptual event. We propose that highly hypnotisable subjects are able to put unusual strong weight on their prior expectations allowing them to counter the weakness of sensory evidence.
Indeed, Hohwy (2013, p. 70) suggests that inter-individual differences in perceptual inference could be explained by how people “set their gain” (p. 70) on their prior expectations versus sensory evidence. For people able to give strong weight to their prior the weakness of the sensory evidence could simply be overcome. In other words, the perceptual decision that there is a mosquito in the room or not, for instance, is essentially determined by the perceptual prior. In contrast, for people not able to give strong weight to their priors perceptual decision is essentially determined by the available sensory evidence (no actual mosquito, no mosquito at all). On this view, the presence of a mental image is necessary but not sufficient to give rise to a hallucinatory-like state. The perceptual prior has to be given a strong weight in order to ignore prediction errors signals arising from the comparison between predicted highly precise sensory evidence and actual weak sensory evidence (Fig 3.).
Lloyd, Lewis, Payne, & Wilson (2011) showed that the number of hallucinatory events participants experienced during perceptual deprivation conditions correlated with their predisposition to hallucinate (r = 0.61) in an everyday context (such as measured by the Revised Hallucination Scale (RHS), Morrison, Wells, & Nothard, 2002). This can be interpreted as inter-individual differences in perceptual inference style: people prone to hallucinate would be people who have a tendency to set the gain on their prior expectations rather on sensory evidence (Hohwy, 2013). Similarly, highly hypnotizable subjects could be people who have a
propensity to set the gain on their prior expectations rather than on sensory evidence.8 If we are right, the hypnotisability score (at least the hallucination item) should correlate with the RHS.
In addition, highly hypnotisable participants should manifest a higher propensity to hallucinate during sensory and perceptual deprivation conditions than medium and low hypnotizable participants. Importantly, we assumed here that the prior produced by the suggestion was a perceptual prior (e.g., there is a fly) but the exact nature of this prior would have to be empirically determined. The prior could equally be conceptualised as an “hyper-prior”, in the sense that it could be more about the nature of mental state (perception vs. imagination) the subject is in than about the world: I am perceiving that there is a fly or I am not imagining that there is a is a fly (what the CCT would call a HOT). In this respect, the propensity to hallucinate might not be higher in highs than in lows or mediums but the level of felt reality attached to their hallucinations or the level of confidence that hallucinations are like perceptions (rather than mere hallucinations) would be higher in highs than in mediums and lows. In addition, if hypnotisability is a function of the gain set on priors, hypnotisability might be (maybe temporarily only) increased after a sensory or perceptual deprivation procedure, to the extent that sensory deprivation conditions ‘force’ the system to rely on priors in order to achieve perceptual inference. Some empirical evidence already supports this latter prediction. Sanders
& Reyher (1969) showed that hypnotisability score was increased (by an average of 4.80 points) after participants had received a sensory deprivation procedure. Finally, if the ability of highs to put strong weight on their prior amounts to a trait ability, we should find differences between highs, mediums and lows outside the hypnotic context in tasks where the level of sensory evidence is varied systematically while the prior is held constant and vice versa. The influence of priors should be higher in highs than in mediums and lows. Thus, highs would need more (and more precise) sensory evidence to update their priors than mediums or lows.
8 For a similar explanation of hallucinations in the context of schizophrenia, see Wilkinson (2014) and Powers, Kelley & Corlett (2016). These authors propose that people with schizophrenia have a bias toward priors.
Halfway between motor suggestions and cognitive suggestions we find challenge suggestions. In challenge suggestions, such as in arm rigidity, there are no movements produced by the participant and the available sensory evidence is low relatively to motor suggestions but there is still some available sensory evidence such as muscular activity signals. During an arm rigidity suggestion we can distinguish different profiles of participants according to the way they conform to the suggestion (Winkel, Younger, Tomcik, Borckardt, & Nash, 2006). Among the participants passing the suggestion, some of them show main activation (as measured by electromyogram or EMG) of both the biceps and triceps thus producing antagonist activity while other participants show main activation of the triceps only. Participants failing the suggestion, bending their arm, show little activation of their triceps (Winkel et al., 2006). We can hypothesise that the arm rigidity suggestion creates the prior that the arm in unbendable (henceforth, unbendable-prior) and that inter-individual differences can still be explained by inter-individual differences in setting the gain on priors. Challenge suggestions are easier than cognitive suggestions, such as hallucination suggestions, because the predictions generated by the prior can be validated by available actual sensory evidence but are more difficult than motor suggestions, such as magnetic hands, because the available sensory is more ambiguous (weak) than in the motor suggestions. The unbendable-prior can generate different sensory predictions, for instance that the triceps and biceps should both activate, as it is the case when we try to bend our arm blocked in something. It can also predict that the triceps only should activate, as it would be the case if our arm were really like an unbendable bar. So, the different strategies employed by the different profiles of participants passing the suggestion (i.e., trying to bend the arm or not trying at all) fit the unbendable-prior well.